Presidential primaries are notoriously hard to forecast. Candidates often announce their runs months (or even years — looking at you, John Delaney) before primary voters make up their minds. The contests are full of unpredictable ups and downs. A bad debate performance can derail a seemingly strong candidate like Marco Rubio, while oddballs like Herman Cain temporarily shoot to the top of the polls. Most reporters and analysts handle this uncertainty in a sensible way: by shrugging, doing their best to cover the candidates fairly and dutifully reminding the public that it’s way too early to know who is going to win the nomination.

But we’re not completely in the dark about who is going to be the Democratic nominee. National primary polls taken at this point in the process don’t have a great track record of picking the eventual winner. But one of the candidates in the top three or four in the national polls often does become the nominee. And while true polling longshots — those polling at zero percent or close to it — do still have some chance of winning, the historical record doesn’t provide much reason for optimism.

The early national polls have a mediocre record of picking the winner at this point in the primary. I’ve analyzed a historical data set of national primary polls from Charles Franklin, the director of the Marquette University Law School Poll, and found that candidates who led a simple average of the last five polls about six months before Iowa won the nomination a little over half of the time. And their predictive power likely won’t increase much over the next few months — polls taken in the fall before the primaries also tend to only pick the winner a bit over half the time. (Note that these results broadly agree with the Economist’s research.)

Joe Biden, the current national poll leader, is arguably weaker than some past front-runners and might have less than a 50 percent win probability. George W. Bush, Al Gore, Bob Dole and Hillary Clinton had all hit 50 percent support in at least one poll by this time in 1999, 1999, 1995 and 2015, respectively, and Biden hasn’t. Unlike Hillary Clinton in 2016 and George W. Bush in 2000, Biden has failed to lock up establishment support or persuade his toughest opponents to stay out of the race. But Biden is still in a decent position — he might follow the path of Mitt Romney, who weathered tough fights with Rick Perry, Newt Gingrich and Rick Santorum, among others, but eventually went on to win the nomination.

And if Biden does falter, the race likely won’t burst wide open: In recent primaries where an early poll leader failed to secure the nomination, the eventual winner tended to still come from the top tier. In the 2008 Democratic primary, the polls put Hillary Clinton ahead. But Barack Obama was consistently in second place in the polls and eventually won the nomination. In the 2008 Republican primary, John McCain hung around third or fourth place for much of the primary season until he won in New Hampshire and eventually claimed the nomination. Before surging in 2004, John Kerry spent 2003 in a multi-way brawl with Dick Gephardt, Howard Dean, Joe Lieberman and (briefly) Wesley Clark.

There are a few cases in which a true underdog won, but those candidates had more support than the lower-tier candidates of 2020 have now. Michael Dukakis was polling in the single digits at this point in the 1988 Democratic primary. In 1991, Bill Clinton seemed stuck in a lower tier with candidates like Jerry Brown and Paul Tsongas, overshadowed by New York governor Mario Cuomo, who never actually entered the race. These examples should be encouraging for Pete Buttigieg or Kamala D. Harris, who have talent and built modest bases but haven’t yet figured out how to vault into the top tier. They still have time.

But this data is worse news for Kirsten Gillibrand, Steve Bullock, Michael Bennet and other seemingly solid candidates who are barely registering in the polls. One of them might break our admittedly short historical patterns and win the nomination. But by this time in the process, the eventual nominee typically managed to register above one percent in the polls.

These numbers don’t rule out a chaotic primary. In 2012, Santorum, Cain and Gingrich all completed at least one surge and decline in the polls between August 2011 and February 2012. But chaos doesn’t always have an impact on the ultimate outcome of a primary. If history is a reliable guide — and it might not be this time — one of the candidates in the top five will likely become the nominee. So Democratic voters overwhelmed by what seems like a plethora of options — as well as party professionals who have influence on debate rules — might want to be realistic. The Democratic field might be crowded, but that hardly means the race is wide open.

Read more: